Remote Sensing of Water and Environment
Chapter 2: Remote Sensing Systems for Earth Observations
Ardeshir Ebtehaj
University of Minnesota
1- Arial Photography
Aerial photography can be characterized as a passive imaging technique operating in the visible and near-infrared parts of the electromagnetic spectrum. Photography is unique among remote sensing systems in that the mechanism for detecting electromagnetic radiation is a photochemical one.
The photographic film constitutes the array of detectors. This consists, in its simplest form, of a suspension of tiny crystals of silver halides in a porous gelatin matrix, supported on a thin plastic film.
Exposure to light can convert a few silver ions in a crystal into metallic silver. The result of this is that the opacity of the exposed and developed film depends in a characteristic way on the amount of light to which it was exposed.
Since the film is more opaque in regions that were more heavily exposed to light, it is termed a negative.
The film as described would respond only to ultraviolet radiation, since only photons of ultraviolet light would be sufficiently energetic to produce the necessary development centers. In practice, the emulsion layer contains sensitizing dyes to extend the spectral response into the visible and sometimes also the near-infrared region.
The lens is at a distance H above the ground, and rays from the ground are brought to a focus in the film plane at distance u beyond the lens. H and u are related to the focal length f of the lens by the following formula:
In practice, since
, it follows that
.
From a consideration of similar triangles it is thus clear that the linear scale s of the negative is given by
.For example, if the negative is 240mm square (this is termed the film’s format) and the camera has a focal length of
mm and the height H is 3000 m, the scale s is 0.00005 (1:20,000) and the coverage is
m. If the spatial resolution on the film is 20 mm, the corresponding resolution at the ground plane is
m. Coverage of larger areas can be obtained through oblique photography, in which the optical axis is not vertical. The principal disadvantage of this approach is that the scale in the ground plane is no longer constant.
1-2 Mapping Relief and Stereophotography
If the ground surface is truly planar and perpendicular to the optical axis, it is clear that the coordinates of objects in this plane can be determined immediately from measurements on the negative once the camera model and scale are known.
However, these conditions will often not be met because of the presence of significant relief in the surface.
The position of a point P in the scene can be expressed by its Cartesian coordinates
. The coordinates of the corresponding point
in the film plane are
. (
elevation of the lens,
elevation of point P) If the object in the scene is vertical, and both its top and its base (on the ground plane) are visible in the image, the above equations can be used twice — once each for the top and the base — to deduce both the x and y coordinates of the object and its height z.
To estimate the relative height of the topgraphy, two vertical photographs are taken of the same scene from different positions. This method is called stereophotography.
To see this formally, we suppose that the first photograph is taken with the geometry shown in the above figure and the second photograph is taken from the same height but with the camera lens shifted by
. These four equations can be solved for x, y, and z to give
The normal arrangement is to choose the baseline such that the overlap is about two thirds of the coverage of a single image.
Photography from spaceborne platforms is rather uncommon, partly because of the difficulty of retrieving exposed film from unmanned spacecraft and partly because of the advent of high-quality electro-optical systems.
The largest publicly available collection of spaceborne photographs consists of nearly 900,000 U.S. military reconnaissance photographs collected by Keyhole cameras on the Corona, Argon, and Lanyard satellite programs between 1959 and 1972. Keyhole-7 satellite reconnaissance photograph showing part of the FranzJosef Land archipelago, 30 April 1965. The area covered by the photograph is approximately 40
22 km. The spatial resolution of the original is about 1 m. Important Note: Since photography is a passive technique operating in visible and near-infrared regions of the electromagnetic spectrum, it can be used only during daylight. Photographs cannot be acquired through clouds.
2- Optical Systems in the Visible, Near- and Short-Infrared Region
Optical systems operating in the visible, near-infrared (VNIR, 400-1100 nm) and short-wave infrared (SWIR, 1100--2500 nm) region possess many similarities to aerial photography. The instruments are largely passive imaging instruments. In this case, electromagnetic radiation is detected electronically, instead of photochemically as with photography. Although VNIR imagers are used both from aircraft and from space, the spaceborne applications tend to dominate.
2-1 Scanning geometry
In pushbroom imaging, the incoming radiation is focused onto a one-dimensional chargecoupled devices (CCDs) array. The instrument thus views instantaneously a narrow strip of the Earth’s surface. Two-dimensional scanning is achieved through the motion of the platform (the aircraft or spacecraft that carries the instrument) relative to the Earth’s surface (e.g., After Landsat 8)
The commonest form of scanning is whiskbroom imaging, though. In this case, a single detector element views the incident radiation. The instantaneous field of view (IFOV) is scanned perpendicularly to the platform motion by a rotating or oscillating mirror within the instrument, while scanning in the forward direction is again achieved through the platform’s forward motion (e.g., befor Landsat 8).
Unlike pushbroom scanners with thusands of detectors, there is only one or a small number of detectors to calibrate on whiskbroom scanners.
2-2 Spatial Resolution
The discrete detector elements or CCDs in a VNIR imager are analogous to the grains in a photographic film, and determine the spatial resolution in a similar manner. The corresponding feature in the image is termed a pixel, or the picture element, and the corresponding feature on the ground is a resolution element.
For example, the ASTER instrument uses a linear CCD array consisting of 5000 detectors, each 7
square. Incoming radiation is projected onto this CCD through optics having a focal length of 329 mm, so the angular width seen by each detector is
microradians. At the nominal observing height of 705 km this corresponds to a horizontal resolution (pixl size) of
m.
2-3 Spectral and Spatial Resolutions
Most VNIR imagers are multispectral, i.e., they provide output in a number of channels or wavebands corresponding to different wavelength ranges. This spectral resolution is normally achieved using filters, which can give bandwidths down to a few nanometers often raning from 20 to 50nm.
Spatial resolutions vary from under a meter to over a kilometer, although the highest spatial resolutions come at the expense of narrow swath widths and long revisit intervals, and have become available only recently.
Like aerial photography, optical sensing is a passive technique, operating in the same regions of the electromagnetic spectrum, and so it can be used only during daylight. Images cannot be acquired through clouds, and cloud cover is a ferquenct in some of the polar regions.
3- Thermal Infrared Systems
Thermal infrared radiation (TIR) is defined as those wavelength between about 8 and 14
. This contains the major part of the black-body radiation emitted by objects at typical terrestrial temperatures, and the principal reason for wanting to detect this radiation is to measure Earth surface temperatures or to deduce surface emissivities. Although the general principles of detection and scanning are similar to those for the VNIR devices, the longer wavelength and hence lower photon energies of TIR radiation do introduce some complications. Thermal infrared imaging is often combined with VNIR imaging in the same instrument.
3-1 Spatial, spectral and radiometric resolution
The factors that determine the spatial resolution of a TIR imager are similar to those for a VNIR imager. The longer wavelength does mean, however, that the spatial resolution is usually coarser. For example, consider the ETM instrument carried on the Landsat 7 satellite. This operates from a nominal height of 705 km, and the optics have an equivalent focal length of 2348 mm. The detectors for the VNIR bands (1–5) have a physical size of 0.1 mm, giving a spatial resolution at the ground of 30 m, while the TIR detector has a size of 0.2 mm, giving a resolution of 60 m. The spectral resolution required for TIR imagers depends on the application. For the Earth surface remote sensing applications, where high spectral resolution is not usually required, the bandwidths in the range 0.1 to 1
are common. Profiling of atmospheric properties requires significantly higher spectral resolution. The wavelengths of these bands are usually around 3 or 4
and around 8 to 14
, thus avoiding the strong water vapor absorption feature around 6 to 7
. Typical noise level in the observed brightness temperatures, often ranges from 0.1 to 1 K, is known as the radiometric resolution. Thermal infrared imaging systems are mainly used for measuring the brightness temperature (TB) of the Earth’s surface and of cloud tops. Thermal infrared imaging systems do not detect reflected sunlight, so they can be used at night. However, the radiation does not penetrate through clouds.
4- Passive Microwave
Passive microwave systems measure radiation in the microwave (wavelengths typically 1 mm to 25 cm, or equivalently frequencies between 1 and 200 GHz) region. As the name implies, it is a passive technique. Since it is a microwave technique, however, it has the ability to penetrate through most clouds. It can thus be characterized as an all-weather, day-and-night, technique. As with thermal infrared systems, the purpose is to measure the brightness temperature of the incident radiation, to deduce either the physical temperature of the Earth’s surface or its emissivity.
The much longer wavelengths of microwave radiation mean that the photons are very much less energetic than those of visible light, so completely different detection techniques are used. A passive microwave radiometer is effectively a radio telescope viewing downward. The incident radiation is collected by an antenna, which converts it into a fluctuating voltage difference that can be amplified and detected.
4-1 Spatial resolution and swath width
The spatial resolution of a passive microwave radiometer is set by the diffraction limit. The angular view of the radiometer is of the order of
, where λ is the wavelength and D is the width (e.g., the diameter in the case of a dish) of the antenna. The long wavelengths imply coarse angular resolutions: for an antenna of 1 m in diameter operating at a wavelength of 2 cm, the angular resolution is of the order of 1 degree (0.02 radian), which would give a horizontal spatial resolution of about
km, from a typical spacecraft altitude of 700 km. This is perhaps the main disadvantage of passive microwave methods. The usual form of mechanical scanning is the conical and cross track scan, in which the beam is rotated in a wide cone about the nadir direction.
Most passive microwave radiometers provide coverage of a number of different frequencies, often in two polarizations. At frequencies below about 20 GHz the detected brightness temperature will be dominated by surface emission. Atmospheric water vapor causes a correction of a few kelvin.
Between about 20 and 35 GHz the surface signal still dominates, although the contribution due to water vapor is significantly larger. Above about 35GHz the effects of molecular absorption in the atmosphere become dominant, and these frequencies are more useful for atmospheric profiling than for surface imaging.
The ‘system noise temperature’ of a few tenths to 1 K are typical, which is a function of the instrument design and its physical temperature, the integration time, and the bandwidth.
Compared with TIR imagers, passive microwave systems provide a substantially poorer spatial resolution but generally a very broad swath width giving synoptic, nearglobal coverage from a few days’ data. Unlike TIR systems they can also be used to observe through most cloud covers.
5- Laser Profiling
Laser profiling is primarily a ranging active sensing method designed to measure the Earth’s surface topography (for example, the surface profile of the ocean or an ice sheet), and therefore not an imaging technique. Laser profilers are usually operated from low-flying aircraft, although satellite instruments do exist. The the Advanced Topographic Laser Altimeter System (ATLAS) is carried on ICESat-2 (launched in 2018). ATLAS has a swath width of 17 m, and is able to measure ice-sheet elevation changes to an accuracy of 4 mm per year. A laser profiler often emits a short pulse of light, usually visible or near-infrared radiation, from a downward-pointing laser (e.g., 532 nm wavelength for ATLAS). The pulse propagates down through the atmosphere, bounces off the Earth’s surface, propagates back up through the atmosphere, and is detected by a photodiode. Therefore, the two-way travel time to the surface can be deduced. If the propagation speed is known, the range to the surface can be determined. If the absolute position of the instrument is known, the absolute position of the reflecting point on the Earth’s surface can therefore also be determined.
The spatial resolution of a laser profiler has two aspects: horizontal resolution and vertical (height) resolution. The horizontal resolution is determined fundamentally by the beam width of the laser. For a laser with a beam width of (radians) operating from a height H above the surface, the horizontal resolution is clearly
For example, a system for which
milliradian operated from
m would illuminate a region on the Earth’s surface of width
m. This is often referred to as the laser’s footprint. The vertical resolution is determined by the accuracy with which the two-way travel time can be measured. The accuracy with which the travel time can be measured is governed by the rise time
of the detected pulse (the time it takes to increase from zero to maximum power) and its signal-to-noise ratio S. For a single pulse, the vertical resolution is given by
where it is assumed that the pulses travel at the speed of light c. Atmospheric constituents affect laser mesurments and thier contamination effects need to be corrected.
6- Radar Altimetry
The radar altimeter is conceptually very similar to the laser profiler. Like the laser profiler, its purpose is to measure the range to the Earth’s surface by measuring the two-way travel time of a short pulse of radiation. The radar altimeters use microwave radiation rather VNIR, with frequencies typically around 10 GHz. This means that, unlike laser profiling, it can observe through cloud. Most other differences between the characteristics of the two types of instrument arise from the much longer wavelength at which the radar altimeter operates.
Spaceborne radar altimetry can achieve vertical resolutions of a few centimeters with a corresponding horizontal resolution of a few kilometers. It is also capable of measuring surface roughness. Corrections are required for observations over sloping terrain, and performance can be significantly degraded where the surface slope exceeds a few degrees or where there are abrupt changes in surface height.


Concep of radar altimetry (left) and SRTM Shaded Relief Anaglyph of Zagros Mountains (right).
7- Radio echo-sounding
This is a technique for measuring the thickness of glaciers and ice sheets. It relies on the fact that freshwater ice is notably transparent to radio frequency radiation in the VHF band (i.e., around 100 MHz or
). Like the two preceding methods, the technique involves transmitting a short pulse of radiation and measuring the two-way travel time for the reflection from the underlying bedrock. Reflections are also obtained from the ice surface and from internal layers if they show sufficiently strong dielectric contrast. Range (thickness) resolutions of the order of 1 m are possible. Because of the long wavelengths of VHF radiation (e.g., 3m in free space at 100 MHz) it is not feasible to construct antennas with narrow beam widths. In order not to compromise the spatial resolution, radio echo-sounding is thus primarily an in situ technique, the equipment being located either on the ice surface itself or in a low-flying aircraft.
Thickness of the Greenland ice sheet derived from radio echo-sounding measurements (Bamber, Layberry, and Gogineni 2001a, b). The data have been gridded at intervals of 5 km and cover an area of 1505
2805 km. 8- Imaging radar and Scatterometry
Imaging radar is a general term to describe active microwave imaging systems. In all of these systems, microwave radiation is beamed toward the Earth’s surface from an antenna, and the scattered radiation is collected by the same antenna and used to build up a two-dimensional picture of the backscattering coefficient of the surface or atmophere. Typical operating frequencies lie between 1 and 10 GHz for the surface and 10 to 100 GHz for the atmopshere. As an active technique, imaging radar is independent of solar illumination and this can operate at night times.
For example, onboard the GPM Core Observatory is the Dual-frequency Precipitation Radar (DPR). The DPR consists of a Ku-band precipitation radar (KuPR) and a Ka-band precipitation radar (KaPR). The KuPR, which operates at 13.6 GHz, is an updated version of the highly successful unit flown on the Tropical Rainfall Measuring Mission (TRMM). The KuPR and the KaPR are co-aligned on the GPM spacecraft such that the 5-km footprint location on the earth is obtained.
8-1 Geometry and Resolution
Radar images can have cross-track or side-looking scanning geometry. For the latter, the antenna does not point vertically downward, but somewhat to one side so that the fan beam illuminates the Earth’s surface only on this side.
In its simplest form, the geometry of imaging radar is illustrated in the following figure.The antenna is long and thin, giving it a ‘‘fan beam’’ region of sensitivity (light shading) that is narrow (angle β) in the direction parallel to the platform motion and broad (angle ψ) in the perpendicular direction.
The radar transmits very short pulses of microwave radiation from the antenna and receive the returned pawer backscattered from the scattering zone -- where it intersects with the object of interest such at the rain drops or the Earth's surface.
For the real aperture radar or side-looking radar, the spatial resolution
in the along-track direction (also called the azimuth direction) is determined by the beam width 
where
is the slanted range of the scatterer,
is the incident angle,
, L is the length of the antenna and λ is the wavelength of the radiation. Therefore,
.Thus fine resolution is achieved by having a long antenna and a small value of the height.
The spatial resolution in the across-track direction (also called the range direction) is determined by the pulse length
, which is the time over which the pulse is emitted, as follows:
.
Because of the dependence of the azimuth resolution on the height, side-looking approach is normally only adopted for airborne systems.
Let us suppose that we wish to achieve a spatial resolution of 10 m, comparable to what can be obtained from VNIR imagers, from a spaceborne imaging radar. If we take nominal values of 5 cm for the wavelength, 800 km for the platform height, and 30 degree for the incidence angle, it can be shown that with a pulse length of 33 nano second or less (which is easy to achieve) , we will need an antenna that is at least 4.6 km long.
This cannot be achieved physically but synthetically through the technique called synthetic aperture radar (SAR). In effect, the data that would have been collected from a very long antenna are synthesized from the data collected as the short antenna is carried along by the platform motion.